Thursday, August 18, 2022

Data is In: Only MOLDX MACs Ever Followed Correct Coding Edit about "1 code for NGS Procedure"

For a number of years, the MolDx program has had billing articles and other instructions that when two or more genes are sequenced together, only 1 unit of 81479 (unlisted code) should be reported, and no code-stacking is allowed.  See the current MolDx article A57880 here.

Several years back, a similar rule crept into the National Correct Coding Initiative (NCCI) edits for Medicare and Medicaid.   This led to a huge pushback from many stakeholders, that "code stacking" in some sequencing situations was fully appropriate, and converting every 2 or more genes into code 81479 (requiring manual processing) would be a nightmare.

By 2020, and it's the same in 2022, CMS had adopted an NCCI rule (Chapter 10 of the correct coding manual) that you were carefully instructed to code any way you wanted.   

Section 8 of Chapter 10, Part F, states that if a lab reports multiple genes with NGS, it should report only ONE unit of service of ONE sequencing code.   If not specific code is available (for example, the genes don't fit the BRCA panel 81432, or don't fit the definition of a tumor panel 81445), then use code 81479.    OR ELSE, they also MAY REPORT stacked sequencing codes that are "medically necessary."   So basically, you can use a panel code (81445) or 81479 or code stacking - dealer's choice, per these multiplex rules.

The data is in for 2020 code usage.  Basically, no MACs except MOLDX MACs used 81479, so we can conclude by inference NGS MAC and FCSO MAC and NOVITAS MAC  weren't trying to apply or enforce a rule that multiple genes should be switched from code stacking to 81479.  It's proof by absence, so to speak.


I've put the spreadsheet of 81479 payments for CY2020 in the cloud here.

Basically, almost none of the 81479 payments, totally $288M dollars, went to labs outside of MolDx jurisdictions.  (The level of exceptions is topped by Interpace, in PA, which got $5M of 81479 payments but likely for specialty testing, not as an alternative to code stacking).  

click to enlarge